On Thu, Jul 22, 2010 at 03:09:03PM -0700, Patrick Pannuto wrote: > This leak is the same as the bus path; cp's kobj name > is set, but kset_register fails, free'ing cp, but not > cp->class_subsys.kobj.name
Same objection here, I don't like how you are having to "know" about how kobject names are handled here.